Hong Kong Speed Figures

[woof43]

For a time I worked with a mob that used Michael Fraser times, and he was going to be paid when that team turned a profit.
The team never turned a profit, and after spending about 7 million(Matthew Tripp financed it) to set it up, folded(I had already walked away thankfully)
Fraser never got paid I don't think(Dom Beirne told me, as he is/was Frasers mate).

I can remember doing a check of Punting form and Accardi(a good few years back), and personally thought they were both rubbish.

I do some verification work for Tasracing, they use that German GPS system, which doesn't work sometimes or intermittently. But I have noted the emails converstions to Tasracing now have Michael Fraser email attached. I'm thinking it must be the same guy, from Horse racing

 

[woof43]

I just remembered something Accardi's mob did with regard to sections in western australia.
They are who provide the sections for their chief racing body(RWWA).
They would use the official 600m time RWWA provided from the lynx camera, and then fit the other sections around that.
How could you trust Accardi accuracy anywhere, if you knew that?

yes some of the camera angles would be just about impossible if you were trying to line up with 200m markers, in the greyhounds you have fixed points around the tracks and fixed cameras and you can create your own timing points, I think Vince does the same, but he would have to make quasi adjustments to get them to 200 markers, just to lineup with official timing, wagering for yourself you don't need to be creating "noise" in your own data which I think is the difference

 

[antonyp22]

I just remembered something Accardi's mob did with regard to sections in western australia.
They are who provide the sections for their chief racing body(RWWA).
They would use the official 600m time RWWA provided from the lynx camera, and then fit the other sections around that.
How could you trust Accardi accuracy anywhere, if you knew that?

Sounds like MF is the best we have then

 

[woof43]

Vic greyhounds is most probably the best for getting accurate data, but even they have FinishLynx at two fixed points which are timing to the nose of every runner. and then have transponders for the other 6 timing points which are measured around the should area as the transponder sit in the rug on their back, so then you have that data being merged and slight adjustments happening and the last sectional is fitted back to the "nose" finish time

 

[seanm8470]

yes some of the camera angles would be just about impossible if you were trying to line up with 200m markers, in the greyhounds you have fixed points around the tracks and fixed cameras and you can create your own timing points, I think Vince does the same, but he would have to make quasi adjustments to get them to 200 markers, just to lineup with official timing, wagering for yourself you don't need to be creating "noise" in your own data which I think is the difference

hey Grant Steve you will find that there are significant differences in the times between Western Australia downloadable CRIS results fille and WA downloadable sectionals file. My western australian database has circa 150K rows of data with both tables of data. it came as a shock to me. on closer inspection the official results had a number of typos plus my excel macro i was using also created some of the drama. WA CRIS they dont make it easy , the stewards report under each race still has the race number tagged on the end lol. the number of macros i needed to design to work around to simply get the data kept me up many nights. i was using ms access x 16 tracks to generate both my localised & universal WA speed ratings. i have now migrated and i use ChatGPT & Grok & a LightGBM 26 feature model.

 

[woof43]

hey Grant Steve you will find that there are significant differences in the times between Western Australia downloadable CRIS results fille and WA downloadable sectionals file. My western australian database has circa 150K rows of data with both tables of data. it came as a shock to me. on closer inspection the official results had a number of typos plus my excel macro i was using also created some of the drama. WA CRIS they dont make it easy , the stewards report under each race still has the race number tagged on the end lol. the number of macros i needed to design to work around to simply get the data kept me up many nights. i was using ms access x 16 tracks to generate both my localised & universal WA speed ratings. i have now migrated and i use ChatGPT & Grok & a LightGBM 26 feature model.

I'm only doing HK racing Sean, I'd never have much confidence in the published sectionals, would rather stick to the greyhounds and HK for an interest. WA seems like a lot of work

 

[StephenJB]

I think it's because WA sections are done by Accardi(a guy called Haydn? Ballantyne did them for him I think) off video.
The 600 was off the lynx camera, where the placement of, was all over the shop I think, and Accardi sections are fitted to that 600 time.
Not sure if that is still the case, but it once was.

I never had many issues with the overall time in Perth(and I used to check them all), but if my memory is holding up, the prov and country were not too good.
Wrong distances were another matter, especially when rail out.

Don't know about others, but I never found a way to win much with sections, not even with the context of overall time.
Not WA, and when I was doing Vic & NSW, I can't remember if there was any sections given out(that was a zillion years back)!

 

[seanm8470]

here are some numbers to give people an idea of some of the results. it is indeed an ongoing work in progress. in my say top 5 features my speed ratings live in the form of derived metrics. btw my top feature is not speed related.

 

[antonyp22]

here are some numbers to give people an idea of some of the results. it is indeed an ongoing work in progress. in my say top 5 features my speed ratings live in the form of derived metrics. btw my top feature is not speed related.

Great results. What do you think is contributing to the adverse results on certain tracks?

 

[seanm8470]

I'm only doing HK racing Sean, I'd never have much confidence in the published sectionals, would rather stick to the greyhounds and HK for an interest. WA seems like a lot of work

hey Granto lol WA has always been hard work. i have always had the belief that my skill set is a transferrable one ..... HK and other jurisdictions are in the pipeline. Cutting my teeth on WA racing augers well. AI simply allows me to churn through more data much more quickly. I still have my core Excel macros i use. i do ponder however whether HK has any value left?. SA & Tas on the surface appeal more to me ie. AI is interesting .... in that it allows much deeper insights .... i recall many years ago i had a light bulb moment and created a speed rating that i felt was different. the raw concept essentially involved "exploding" the ratings. i picked it up from a maths professor in the USA. i was always fascinated by the way he simply looked at numbers. complex maths is a bit over my head. because i have heaps of time on my hands lol , recently i revisited the construct and pulled it apart and back to the universal method it was derived from. AI said i was doing it wrong and my way would not hold up. surprise surprise. AI got it wrong. but it gave me a platform that i could use to quickly test it. didn't mean to chew your ear off. just waiting for some code to finish.

 

[seanm8470]

Great results. What do you think is contributing to the adverse results on certain tracks?

the quality of the underlying data for one. this can be diminished somewhat simply through the ratings method being used and the logic to arrive at a number. good coverage can reduce the impact of poor data. inter track runs in the country .provincial areas in particular will be short as i don/t have the sectional data for say places like Mt Magnet , Collie etc etc therefore the data will simply be incomplete. as i have said beforehand , this is a work in progress. but as you can see there is plenty of merit in those numbers. also still getting my head around the AI side of things.

 

[The Hare]

Card for Sunday with another Group 2 handicap and latest results.

 

[desimal]

here are some numbers to give people an idea of some of the results. it is indeed an ongoing work in progress. in my say top 5 features my speed ratings live in the form of derived metrics. btw my top feature is not speed related.

How are you betting here, early fixed odds with MBLs or is there enough liquidity (and inefficiency) on Betfair for smaller WA tracks, or is it mostly smart money there?

i do ponder however whether HK has any value left?

Very hard competing on tote without rebates and exotic bets. But it's some of the best racing out there and most fun doing work on.

I live in Japan but from Aus(WA) and betting JPN/HK/AUS, and I think people in Aus sometimes don't realise how lucky they are to have fixed odd MBLs and particularly Betfair where you are assured a price and low % markets, vs overseas volatile odds where teams somewhat dictate closing tote odds.

Better racing product in Asia (imo) but tough markets for betting.

On a side note does anyone know if the Ladbrokes machines in Queensland offer fixed odds for Hong Kong races? That would be a dream for HK bettors, but I imagine they would still have ways to ban you...